Should All These Searches Be Forgotten?

The big search engines released end-of-the-year lists that intrigue and entertain. They can also remind us of the fundamentals of search user behavior. Each year when we look back at top searches, we get to see what searchers are up to and measure the strides in human development over the last twelve months. Our curiosity. Our passion. Etched for eternity in the eloquent keywords used by millions demanding cars and jennifer lopez.

Even with the tens of thousands of search sessions I’ve witnessed, I still get a pang of, “Really?! People searched that? Is this stuff real?”

All those searches I’ve seen tell me these lists are accurate snapshots coming from the engines. Granted, I think they are cleaned up–people don’t capitalize much, and don’t often spell out Hillary Rodham Clinton (and probably don’t get the spelling right all the time when they do)–but this does show the short head in all its glory. The filtering makes sense too–focusing on “fastest-rising” and “Top Ten Send-offs” takes us down the long tail a bit, and makes for more entertaining releases.

The basic navigational searches, where users are typing in the name of a site on a search engine, tell us a lot about searcher behavior. They underscore that there’s a whole lotta satisficing going on.

Gord’s explanation of “satisficing” in a previous Just Behave column is spot-on. Users don’t optimize from a careful consideration of all choices in most decision making, they take the first plausible route. This applies not only to their behavior on SERPs, but to everything else they do: navigating the Web, conceiving search topics, typing keywords, etc. People do constantly type myspace and google and ebay into search engines. I would even bet google is persistently a top search on Google.

So, what’s going on here?

In user testing over the last 12 years, I’ve seen this behavior many times. At first, we were a little shocked. When we talked with people, it made sense, for a variety of reasons.

The main thing that influences them to type in domain names to navigate is that it works. For example, I’ve met many users who have an ISP start page set up as their home page. It’s just been that way since they got the Internet connection. That page features a Web search box. So when someone says “you should check out MySpace,” they will type that into their search box (their daily on ramp to the rest of the Web), and arrive at MySpace after a click on the SERP. Next time they need to get to MySpace, it’s motor memory. Of course, there are many users who would bookmark MySpace or use the browser’s URL field to type in “www.myspace.com,” but you can see how navigating via the search box might be easier.

Another context in play here is the frequency of search engines being on screen. If a user wants to navigate to a domain, a search box is often the closest, simplest interface to get there. Fitts’ Law tells us that it’s easier to move a cursor to a target that’s less distance away and larger than another target. If someone’s on a search engine, that site’s search box is often faster to get to than the browser’s URL field.

A third class of behavior here would be informational searches where we’d see engagement with more of the choices on a SERP. Think of someone learning about Facebook for the first time and trying to get an overview of what it is. Another user already knowing what Facebook is and hearing about it in the news might just type in the word–trusting the search engine to display news results in the mix.

It’s as if users have come to learn that search is aware of the same context they are. If you mentioned “iPhone” to a friend the day it came out, they’d be aware of the same buzz as you, and would be interested in reviews, pictures, news about availability, etc. A few months later, the expectation if you mentioned “iPhone” to your friend would be that she knew that you wanted to chat about the price drop and what those first day purchasers thought. In both cases, “iphone” would be the search recorded, but the search results would be tuned to the contemporary context.

3 Engines, 3 insights Into the same behavior

This year Ask came up with some fun categories, such as “Top Celebrity Search of Pregnant Stars in 2007” and “Top Presidential Candidate Searches of 2007,” in addition to the straight up searches featuring “dictionary” and “movies.” Google focused on the fastest rising queries they saw in 2007, and Yahoo on the top trends in search in 2007. Note all the very simple one and two word searches.

A good illustration of users boiling down the essence of a concept to a few characters but expecting the search engine to know what they want is described in Marissa Mayer’s 2007 Google Zeitgeist webcast. This look at Google Trends shows jumps in the simple search math following news stories about the academic subject. Two stories came out–the first discussing a poll that showed math was the least popular subject at school, and the second that SAT math scores had reached a new high.

My take on this is that user curiosity is piqued by the stories–including hearing about them verbally from others–and they go to a search engine and drop in the shortest query possible to get the scoop. The engines will look at their current news feeds, see many matches for “math,” and return news articles to the top of the page. Everyone’s happy. The point is that no one wanted to (or had to) enter recent high school subject popularity poll math.

It’s not just satisficing driving this–it’s previous experience with search engines rising to meet needs for short queries. The engines have learned how to balance the possible meanings and blend them into the results. That, of course, has led to more confidence in drawing together various structured and unstructured data; hence, Ask 3D and Google Universal.

With many of these searches, such as those for Britney (and yes, try just putting britney in; the engines make an educated guess who you mean), the blend of results coming back need to cover all kinds of user needs. To get a peek at the needs that aren’t (yet) expressed on those basic queries, look to the various suggestions and related search products each engine features. Yahoo’s Search Assistant and Ask’s TypeAhead Suggestions and Zoom Related Search appear on their main search interfaces. Google Suggest is in their labs area, and their various development flavors of related search appear on small samples of their traffic. These are some if the semantic maps people hold–the variety of implicit keywords behind the short queries. Their utility is validated on Ask: more than 20% of the time users take advantage of Zoom Related Search to refocus the results–it’s the most used search tool on Ask.com.

When we see users searching topics as generic as cars, it may be tempting to think “Wow, people are really unable to articulate what they want.” Search engines realized long ago that they have to meet the user more than half way. In the case of cars , the searcher’s semantic map varies from user to user. One might be at the start of the buying cycle and expect to see buying guides. Another may want access to a list of online used car dealers. Another may want to understand how cars work. And another may want to buy the Pixar film’s DVD.

Since relevance is in the mind of the beholder, good SERPs cover many different needs at once. The next time this user returns, they may drop movies in the search box to get a list of what’s currently out. Or the Top 100 of all time. Or the history of film. Little effort on the user’s part, listening and blending on the engine’s part.

Search marketers can learn from these is several ways:

• The most simple terms will often collect the most honey (though, as Gord points out, the conversion rate will be lower than those down the long tail.)
• Users cast a wide net to start with, and then narrow down. Today’s SERPs reflect this more and more.
• Keep up with news and buzz in your category–these will create new keywords and combinations that can result in traffic.
• People are interested in other people–many of these top searches have to do with people and communication. The iPhone is a communication device AND a cultural phenomenon that people want to be informed about.
• Context can change for the same keywords–for one user over time, and from user to user.

When search engines talk about Top 10 lists, it might seem like there’s nothing to be learned about the keywords and clients you represent–that these are just ephemeral human-interest stories. Actually, these searches tell a story about user behavior that will be with us for a long time. To the public, these list are entertaining. To the search industry, they provide critical insight.

Michael Ferguson is senior user experience analyst for Ask.com. The Just Behave column appears Fridays at Search Engine Land.